Impact of Population Growth

Author(s): Paul R. Ehrlich and John P. Holdren

Source: Science, New Series, Vol. 171, No. 3977 (Mar. 26, 1971), pp. 1212-1217
Published by: American Association for the Advancement of Science
Stable URL: http://www.jstor.org/stable/1731166 .
Accessed: 24/08/2011 15:29

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and
extend access to Science.

http://www.jstor.org
 resource demand or environmental im-
pact for absolute per capita increases,
and thus to underestimate the role of
the population multiplier. Moreover,
it is often assumed that population size
Impact of Population Growth and per capita impact are independent
variables, when in fact they are not.
Consider, for example, the recent article
Complacency concerning this component of man's by Coale (1), in which he disparages
the role of U.S. population growth in
predicament is unjustified and counterproductive. environmental problems by noting that
since 1940 "population has increased
Paul R. Ehrlich and John P. Holdren by 50 percent, but per capita use of
electricity has been multiplied several
times." This argument contains both the
fallacies to which we have just re-
ferred.
The interlocking crises in popula- physical environment of urban ghettos, First, a closer examination of very
tion, resources, and environment have the human behavioral environment, rapid increases in many kinds of con-
been the focus of countless papers, and the epidemiological environment. sumption shows that these changes re-
dozens of prestigious symposia, and a 5) Theoretical solutions to our prob- flect a shift among alternatives within
growing avalanche of books. In this lems are often not operational and a larger (and much more slowly grow-
wealth of material, several questionable sometimes are not solutions.
ing) category. Thus the 760 percent
assertions have been appearing with We now examine these theorems increase in electricity consumption
increasing frequency. Perhaps the most in some detail. from 1940 to 1969 (4) occurred in
serious of these is the notion that the large part because the electrical com-
size and growth rate of the U.S. popu- ponent of the energy budget was (and
lation are only minor contributors to Population
Size and
is) increasing much faster than the
this country's adverse impact on local Per Capita Impact budget itself. (Electricity comprised 12
and global environments (1, 2). We percent of the U.S. energy consump-
propose to deal with this and several In an agricultural or technological tion in 1940 versus 22 percent today.)
related misconceptions here, before society, each human individual has a The total energy use, a more important
persistent and unrebutted repetition negative impact on his environment. figure than its electrical component in
entrenches them in the public mind- He is responsible for some of the sim- terms of resources and the environ-
if not the scientific literature. Our dis- plification (and resulting destabiliza- ment, increased much less dramatical-
cussion centers around five theorems tion) of ecological systems which re- ly-140 percent from 1940 to 1969.
which we believe are demonstrably sults from the practice of agriculture Under the simplest assumption (that
true and which provide a framework (3). He also participates in the utili- is, that a given increase in population
for realistic analysis: zation of renewable and nonrenewable size accounts for an exactly propor-
1) Population growth causes a dis- resources. The total negative impact tional increase in consumption), this
proportionate negative impact on the of such a society on the environment would mean that 38 percent of the
environment. can be expressed, in the simplest terms, increase in energy use during this
pe-
2) Problems of population size and by the relation riod is explained by population growth
growth, resource utilization and deple- I=P . F (the actual population increase from
tion, and environmental deterioration 1940 to 1969 was 53 percent). Similar
must be considered jointly and on a where P is the population, and F is a considerations reveal the imprudence
global basis. In this context, popula- function which measures the per capita of citing, say, aluminum consumption
tion control is obviously not a panacea impact. A great deal of complexity is to show that population growth is an
-it is necessary but not alone sufficient subsumed in this simple relation, how- "unimportant" factor in resource use.
to see us through the crisis. ever. For example, F increases with Certainly, aluminum consumption has
3) Population density is a poor mea- per capita consumption if technology swelled by over 1400 percent since
sure of populaction pressure, and re- is held constant, but may decrease in 1940, but much of the increase has
distributing population would be a some cases if more benign technol- been due to the substitution of aluim-
dangerous pseudosolution to the pop- ogies are introduced in the provision inum for steel in many applications.
ulation problem. of a constant level of consumption. Thus a fairer measure is combined
4) "Environment" must be broadly (We shall see in connection with theo- consumption of aluminum and steel,
construed to include such things as the rem 5 that there are limits to the im- which has risen only 117 percent since
provements one should anticipate from 1940. Again, under the simplest as-
Dr. Ehrlich is professor of biology at Stanford such "technological fixes.") sumption, population growth accounts
University, Palo Alto, California, and Dr. Holdren Pitfalls abound in the interpretation for 45 percent of the increase.
is a physicist at the Lawrence Radiation Labora-
tory, University of California, Livermore. This of manifest increases in the total im- The "simplest assumption" is not
article is adapted from a paper presented before
the President's Commission on Population Growth pact I. For instance, it is easy to mis- valid, however, and this is the second
and the American Future on 17 November 1970. take changes in the composition of flaw in Coale's example (and in his
1212 SCIENCE, VOL. 171
thesis). In short, he has failed to rec-
ognize that per capita consumption of
energy and resources, and the asso-
ciated per capita impact on the envi-
ronment, are themselves functions of
the population size. Our previous equa-
tion is more accurately written
-=P F (P)
displaying the fact that impact can in-
crease faster than linearly with popu-
lation. Of course, whether F (P) is an
increasing or decreasing function of P
depends in part on whether diminishing
returns or economies of scale are
dominant in the activities of im-
portance. In populous, industrial na-
tions such as the United States, most
economies of scale are already being
exploited; we are on the diminishing
returns part of most of the important
curves.
As one example of diminishing re-
turns, consider the problem of provid-
ing nonrenewable resources such as
minerals and fossil fuels to a growing
population, even at fixed levels of pei
capita consumption. As the richest sup-
plies of these resources and those near-
est to centers of use are consumed,
we are obliged to use lower-grade ores,
drill deeper, and extend our supply
networks. All these activities increase
our per capita use of energy and our
per capita impact on the environment.
In the case of partly renewable re-
sources such as water (which is ef-
fectively nonrenewable when ground-
water supplies are mined at rates far
exceeding natural recharge), per capita
costs and environmental impact esca-
late dramatically when the human
population demands more than is
locally available. Here the loss of free-
flowing rivers and other economic,
esthetic, and ecological costs of mas-
sive water-movement projects repre-
sent'increased per capita diseconomies
directly stimulated by population
growth.
Diminishing returns are also opera-
tive in increasing food production to
meet the needs of growing populations.
Typically, attempts are made both to
overproduce on land already farmed
and to extend agriculture to marginal
land. The former requires dispropor-
tionate energy use in obtaining and dis-
tributing water, fertilizer, and pesti-
cides. The latter also increases per
capita energy use, since the amount of
energy invested per unit yield increases
as less desirable land is cultivated.
26 MARCH 1971
Similarly, as the richest fisheries stocks
are depleted, the yield per unit effort
drops, and more and more energy per
capita is required to maintain the sup-
ply (5). Once a stock is depleted it
may not recover-it may be nonre-
newable.
Population size influences per capita
impact in ways other than diminishing
returns. As one example, consider the
oversimplified but instructive situation
in which each person in the popula-
tion has links with every other person
-roads, telephone lines, and so forth.
These links involve energy and ma-
terials in their construction and use.
Since the number of links increases
much more rapidly than the number
of people (6), so does the per capita
consumption associated with the links.
Other factors may cause much
steeper positive slopes in the per capita
impact function, F (P). One such
phenomenon is the threshold effect. Be-
low a certain level of pollution trees
will survive in smog. But, at some
point, when a small increment in
population produces a small increment
in smog, living trees become dead
trees. Five hundred people may be
able to live around a lake and dump
their raw sewage into the lake, and the
natural systems of the lake will be
able to break down the sewage and
keep the lake from undergoing rapid
ecological change. Five hundred and
five people may overload the system
and result in a "polluted" or eutrophic
lake. Another phenomenon capable of
causing near-discontinuities is the
synergism. For instance, as cities push
out into farmland, air pollution in-
creasingly becomes a mixture of agri-
cultural chemicals with power plant
and automobile effluents. Sulfur diox-
ide from the city paralyzes the clean-
ing mechanisms of the lungs, thus in-
creasing the residence time of potential
carcinogens in the agricultural chemi-
cals. The joint effect may be much
more than the sum of the individual
effects. Investigation of synergistic ef-
fects is one of the most neglected areas
of environmental evaluation.
Not only is there a connection be-
tween population size and per capita
damage to the environment, but the
cost of maintaining environmental
quality at a _given level escalates dis-
proportionately as population size in-
creases. This effect occurs in part be-
cause costs increase very rapidly as one
tries to reduce contaminants per unit
volume of effluent to lower and lower
levels (diminishing returns again!).
Consider municipal sewage, for ex-
ample. The cost of removing 80 to 90
percent of the biochemical and chem-
ical oxygen demand, 90 percent of the
suspended solids, and 60 percent of the
resistant organic material by means of
secondary treatment is about 8 cents
per 1000 gallons (3785 liters) in a large
plant (7). But if the volume of sewage is
such that its nutrient content creates a
serious eutrophication problem (as is
the case in the United States today),
or if supply considerations dictate the
reuse of sewage water for industry,
agriculture, or groundwater recharge,
advanced treatment is necessary. The
cost ranges from two to four times
as much as for secondary treatment
(17 cents per 1000 gallons for carbon
absorption; 34 cents per 1000 gallons
for disinfection to yield a potable sup-
ply). This dramatic example of dimin-
ishing returns in pollution control
could be repeated for stack gases, auto-
mobile exhausts, and so forth.
Now consider a situation in which
the limited capacity of the environ-
ment to absorb abuse requires that we
hold man's impact in some sector con-
stant as population doubles. This means
per capita effectiveness of pollution
control in this sector must double
(that is, effluent per person must be
halved). In a typical situation, this
would yield doubled per capita costs,
or quadrupled total costs (and proba-
bly energy consumption) in this sector
for a doubling of population. Of course,
diminishing returns and threshold ef-
fects may be still more serious: we
may easily have an eightfold increase
in control costs for a doubling of popu-
lation. Such arguments leave little
ground for the assumption, popularized
by Barry Commoner (2, 8) and others,
that a 1 percent rate of population
growth spawns only 1 percent effects.
It is to be emphasized that the pos-
sible existence of "economies of scale"
does not invalidate these arguments.
Such savings, if available at all, would
apply in the case of our sewage ex-
ample to a change in the amount of
effluent to be handled at an installation
of a given type. For most technologies,
the United States is already more than
populous enough to achieve such econ-
omies and is doing so. They are ac-
counted for in our example by citing
figures for the largest treatment plants
of each type. Population growth, on
1213
 the other hand, forces us into quanti-
tative and qualitative changes in how
we handle each unit volume of efflu-
ent-what fraction and what kinds of
material we remove. Here economies
of scale do not apply at all, and dimin-
ishing returns are the rule.
Global Context
We will not deal in detail with the
best example of the global nature and
interconnections of population resource
and environmental problems-namely,
the problems involved in feeding a
world in which 10 to 20 million peo-
ple starve to death annually (9), and
in which the population is growing by
some 70 million people per year. The
ecological problems created by high-
yield agriculture are awesome (3, 10)
and are bound to have a negative feed-
back on food production. Indeed, the
Food and Agriculture Organization of
the United Nations has reported that
in 1969 the world suffered its first
absolute decline in fisheries yield since
1950. It seems likely that part of this
decline is attributable to pollution
originating in terrestrial agriculture.
A second source of the fisheries de-
cline is, of course, overexploitation of
fisheries by the developed countries.
This problem, in turn, is illustrative
of the situation in regard to many
other resources, where similarly rapa-
cious and shortsighted behavior by the
developed nations is compromising the
aspirations of the bulk of humanity to
a decent existence. It is now becoming
more widely comprehended that the
United States alone accounts for per-
haps 30 percent of the nonrenewable
resources consumed in the world each
year (for example, 37 percent of the
energy, 25 percent of the steel, 28 per-
cent of the tin, and 33 percent of the
synthetic rubber) (11). This behavior
is in large part inconsistent with Amer-
ican rhetoric about "developing" the
countries of the Third World. We may
be able to afford the technology to
mine lower grade deposits when we
have squandered the world's rich ores,
but the underdeveloped countries, as
their needs grow and their means re-
main meager, will not be able to do so.
Some observers argue that the poor
countries are today economically de-
pendent on our use of their resources,
and indeed that economists in these
countries complain that world demand
for their raw materials is too low (1).
1214
This proves only that their economists ports 63 percent of its cereals, in-
are as shortsighted as ours. cluding 100 percent of its corn and
It is abundantly clear that the entire rice. It also imports all of its cotton,
context in which we view the world 77 percent of its wool, and all of its
resource pool and the relationships be- iron ore, antimony, bauxite, chromium,
tween developed and underdeveloped copper, gold, lead, magnesite, manga-
countries must be changed, if we are to nese, mercury, molybdenum, nickel,
have any hope of achieving a stable silver, tin, tungsten, vanadium, zinc,
and prosperous existence for all human phosphate rock (fertilizer), potash
beings. It cannot be stated too force- (fertilizer), asbestos, and diamonds. It
fully that the developed countries (or, produces energy equivalent to some
more accurately, the overdeveloped 20 million metric tons of coal and
countries) are the principal culprits in consumes the equivalent of over 47
the consumption and dispersion of the million metric tons (14).
world's nonrenewable resources (12) A certain preoccupation with density
as well as in appropriating much more as a useful measure of overpopulation
than their share of the world's protein. is apparent in the article by Coale (1).
Because of this consumption, and be- He points to the existence of urban
cause of the enormous negative impact problems such as smog in Sydney,
on the global environment accompany- Australia, "even though the total popu-
ing it, the population growth in these lation of Australia is about 12 million
countries must be regarded as the most in an area 80 percent as big as the
serious in the world today. United States," as evidence that en-
In relation to theorem 2 we must vironmental problems are unrelated to
emphasize that, even if population population size. His argument would
growth were halted, the present popu- be more persuasive if problems of
lation of the world could easily destroy population distribution were the only
civilization as we know it. There is a ones with environmental consequences,
wide choice of weapons-from un- and if population distribution were un-
stable plant monocultures and agri- related to resource distribution and
cultural hazes to DDT, mercury, and population size. Actually, since the
thermonuclear bombs. If population carrying capacity of the Australian con-
size were reduced and per capita con- tinent is far below that of the United
sumption remained the same (or in- States, one would expect distribution
creased), we would still quickly run problems-of which Sydney's smog is
out of vital, high-grade resources or one symptom-to be encountered at a
generate conflicts over diminishing sup- much lower total population there. Re-
plies. Racism, economic exploitation, sources, such as water, are in very
and war will not be eliminated by short supply, and people cluster where
population control (of course, they are resources are available. (Evidently, it
unlikely to be eliminated without it). cannot be emphasized enough that
carrying capacity includes the avail-
ability of a wide variety of resources
Population Density and Distribution in addition to space itself, and that
population pressure is measured rela-
Theorem 3 deals with a problem re- tive to the carrying capacity. One would
lated to the inequitable utilization of expect water, soils, or the ability of
world resources. One of the common- the environment to absorb wastes to
est errors made by the uninitiated is be the limiting resource in far more
to assume that population density (peo- instances than land area.)
ple per square mile) is the critical mea- In addition, of course, many of the
sure of overpopulation or underpopu- most serious environmental problems
lation. For instance, Wattenberg states are essentially independent of the way
that the United States is not very in which population is distributed.
crowded by "international standards" These include the global problems of
because Holland has 18 times the popu- weather modification by carbon dioxide
lation density (13). We call this no- and particulate pollution, and the
tion "the Netherlands fallacy." The threats to the biosphere posed by man's
Netherlands actually requires large massive inputs of pesticides, heavy
chunks of the earth's resources and metals, and oil (15). Similarly, the
vast areas of land not within its borders problems of resource depletion and
to maintain itself. For example, it is ecosystem simplification by agriculture
the second largest per capita importer depend on how many people there are
of protein in the world, and it im- and their patterns of consumption, but
SCIENCE, VOL. 171
not in any major way on how they are
distributed.
Naturally, we do not dispute that
smog and most other familiar urban
ills are serious problems, or that they
are related to population distribution.
Like many of the difficulties we face,
these problems will not be cured sim-
ply by stopping population growth; di-
rect and well-conceived assaults on the
problems themselves will also be re-
quired. Such measures may occasion-
ally include the redistributionof popu-
lation, but the considerable difficulties
and costs of this approach should not
be underestimated. People live where
they do not because of a perverse in-
tention to add to the problems of their
society but for reasons of economic
necessity, convenience, and desire for
agreeable surroundings.Areas that are
uninhabited or sparsely populated to-
day are presumably that way because
they are deficient in some of the requi-
site factors. In many cases, the remedy,
for such deficiencies-for example, the
provision of water and power to the
wastelands of central Nevada-would
be extraordinarilyexpensive in dollars,
energy, and resources and would prob-
ably create environmentalhavoc. (Will
we justify the rape of Canada's rivers
to "colonize" more of our western
deserts?)
Moving people to more "habitable"
areas, such as the central valley of
California or, indeed, most suburbs,
exacerbates another serious problem-
the paving-over of prime farmland.
This is already so serious in California
that, if current trends continue, about
50 percent of the best acreage in the
nation's leading agricultural state will
be destroyed by the year 2020 (16).
Encouraging that trend hardly seems
wise.
Whatever attempts may be made to
solve distribution-related problems,
they will be undermined if population
growth continues, for two reasons.
First, population growth and the ag-
gravation of distribution problems are
correlated-part of the increase will
surely be absorbed in urban areas that
can least afford the growth. Indeed,
barringthe unlikely prompt reversal of
present trends, most of it will be ab-
sorbedthere. Second, populationgrowth
puts a disproportionate drain on the
very financial resourcesneeded to com-
bat its symptoms. Economist Joseph
Spengler has estimated that 4 percent
of national income goes to support our
1 percent per year rate of population
26 MARCH 1971
growth in the United States (17). The cine has made widespread plague im-
4 percent figure now amounts to about possible (21). The Asian influenza
$30 billion per year. It seems safe to epidemic of 1968 killed relatively few
conclude that the faster we grow the people only because the virus hap-
less likely it is that we will find the pened to be nonfatal to people in other-
funds either to alter population distribu- wise good health, not because of pub-
tion patterns or to deal more compre- lic health measures. Far deadlier
hensively and realistically with our viruses, which easily could be scourges
problems. without precedent in the population
at large, have on more than one oc-
casion been confined to research
Meaning of Environment workers largely by good luck [for
example, the Marburgvirusincident of
Theorem 4 emphasizes the compre- 1967 (22) and the Lassa fever incident
hensiveness of the environment crisis. of 1970 (21, 23)].
All too many people think in terms of
national parks and trout streams when
they say "environment."For this rea- Solutions: Theoretical and Practical
son many of the suppressedpeople of
our nation consider ecology to be just Theorem 5 states that theoretical
one more "racist shuck" (18). They solutions to our problems are often not
are apathetic or even hostile toward operational, and sometimes are not
efforts to avert further environmental solutions. In terms of the problem of
and sociological deterioration, because feeding the world, for example, tech-
they have no reason to believe they nological fixes suffer from limitations
will share the fruits of success (19). in scale, lead time, and cost (24). Thus
Slums, cockroaches, and rats are eco- potentially attractive theoretical ap-
logical problems, too. The correction proaches-such as desalting seawater
of ghetto conditions in Detroit is for agriculture,new irrigation systems,
neither more nor less important than high-protein diet supplements-prove
saving the Great Lakes-both are im- inadequate in practice. They are too
perative. little, too late, and too expensive, or
We must pay careful attention to they have sociological costs which
sources of conflict both within the hobble their effectiveness (25). More-
United States and between nations. over, many aspects of our technological
Conflict within the United States blocks fixes, such as synthetic organic pesti-
progress toward solving our problems; cides and inorganic nitrogen fertilizers,
conflict among nations can easily have created vast environmentalprob-
"solve" them once and for all. Recent lems which seem certain to erode global
laboratory studies on human beings productivity and ecosystem stability
support the anecdotal evidence that (26). This is not to say that important
crowding may increase aggressiveness gains have not been made through the
in human males (20). These results application of technology to agricul-
underscore long-standing suspicions ture in the poor countries, or that
that population growth, translated further technological advances are not
through the inevitable uneven distribu- worth seeking. But it must be stressed
tion into physical crowding, will tend that even the most enlightened tech-
to make the solution of all of our nology cannot relieve the necessity of
problems more difficult. grappling forthrightly and promptly
As a final example of the need to with population growth [as Norman
view "environment"broadly, note that Borlaug aptly observedon being notified
human beings live in an epidemiologi- of his Nobel Prize for development of
cal environment which deteriorates the new wheats (27)].
with crowding and malnutrition-both Technological attempts to ameliorate
of which increase with population the environmental impact of popula-
growth.The hazardposed by the preva- tion growth and rising per capita afflu-
lence of these conditions in the world ence in the developed countries suffer
today is compounded by man's un- from practical limitations similar to
precedented mobility: potential car- those just mentioned. Not only do such
riers of diseases of every description measures tend to be slow, costly, and
move routinely and in substantial insufficientin scale, but in additionthey
numbers from continent to continent most often shift our impact rather than
in a matter of hours. Nor is there any remove it. For example, our first
reason to believe that modern medi- generation of smog-control devices in-
1215
 creased emissions of oxides of nitrogen
while reducing those of hydrocarbons
and carbon monoxide. Our unhappiness
about eutrophication has led to the
replacement of phosphates in deter-
gents with compounds like NTA-
nitrilotriacetic acid-which has car-
cinogenic breakdown products and ap-
parently enhances teratogenic effects
of heavy metals (28). And our distaste
for lung diseases apparently induced
by sulfur dioxide inclines us to ac-
cept the hazards of radioactive waste
disposal, fuel reprocessing, routine
low-level emissions of radiation, and
an apparently small but finite risk of
catastrophic accidents associated with
nuclear fission power plants. Similarly,
electric automobiles would simply shift
part of the environmental burden of
personal transportation from the vicin-
ity of highways to the vicinity of power
plants.
We are not suggesting here that elec-
tric cars, or nuclear power plants, or
substitutes for phosphates are inher-
ently bad. We argue rather that they,
too, pose environmental costs which
must be weighed against those they
eliminate. In many cases the choice is
not obvious, and in all cases there will
be some environmental impact. The
residual per capita impact, after all the
best choices have been made, must then
be multiplied by the population en-
gaging in the activity. If there are too
many people, even the most wisely
managed technology will not keep the
environment from being overstressed.
In contending that a change in the
way we use technology will invalidate
these arguments, Commoner (2, 8)
claims that our important environ-
mental problems began in the 1940's
with the introduction and rapid spread
of certain "synthetic" technologies:
pesticides and herbicides, inorganic
fertilizers, plastics, nuclear energy, and
high-compression gasoline engines. In
so arguing, he appears to make two
unfounded assumptions. The first is
that man's pre-1940 environmental im-
pact was innocuous and, without
changes for the worse in technology,
would have remained innocuous even at
a much larger population size. The
second assumption is that the advent
of the new technologies was indepen-
dent of the attempt to meet human
needs and desires in a growing popu-
lation. Actually, man's record as a
simplifier of ecosystems and plunderer
of resources can be traced from his
1216
probable role in the extinction of many
Pleistocene mammals (29), through the
destruction of the soils of Mesopotamia
by salination and erosion, to the de-
forestation of Europe in the Middle
Ages and the American dustbowls of
the 1930's, to cite only some highlights.
Man's contemporary arsenal of syn-
thetic technological bludgeons indis-
putably magnifies the potential for
disaster, but these were evolved in
some measure to cope with popula-
tion pressures, not independently of
them. Moreover, it is worth noting that,
of the four environmental threats
viewed by the prestigious Williamstown
study (15) as globally significant, three
are associated with pre-1940 tech-
nologies which have simply increased
in scale [heavy metals, oil in the seas,
and carbon dioxide and particulates in
the atmosphere, the latter probably due
in considerable part to agriculture
(30)]. Surely, then, we can anticipate
that supplying food, fiber, and metals
for a population even larger than
today's will have a profound (and
destabilizing) effect on the global
ecosystem under any set of technolog-
ical assumptions.
Conclusion
John Platt has aptly described man's
present predicament as "a storm of
crisis problems" (31). Complacency
concerning any component of these
problems-sociological, technological,
economic, ecological-is unjustified and
counterproductive. It is time to admit
that there are no monolithic solutions
to the problems we face. Indeed, pop-
ulation control, the redirection of tech-
nology, the transition from open to
closed resource cycles, the equitable
distribution of opportunity and the
ingredients of prosperity must all be
accomplished if there is to be a future
worth having. Failure in any of these
areas will surely sabotage the entire
enterprise.
In connection with the five theorems
elaborated here, we have dealt at length
with the notion that population
growth in industrial nations such as the
United States is a minor factor, safely
ignored. Those who so argue often
add that, anyway, population control
would be the slowest to take effect of
all possible attacks on our various
problems, since the inertia in attitudes
and in the age structure of the popu-
lation is so considerable. To conclude
that this means population control
should be assigned low priority strikes
us as curious logic. Precisely because
population is the most difficult and
slowest to yield among the components
of environmental deterioration, we
must start on it at once. To ignore
population today because the problem
is a tough one is to commit ourselves
to even gloomier prospects 20 years
hence, when most of the "easy" means
to reduce per capita impact on the en-
vironment will have been exhausted.
The desperate and repressive measures
for population control which might be
contemplated then are reason in them-
selves to proceed with foresight,
alacrity, and compassion today.
References and Notes
1. A. J. Coale, Science 170, 132 (1970).
2. B. Commoner, Saturday Rev. 53, 50 (1970);
Humanist 30, 10 (1970).
3. For a general discussion, see P. R. Ehrlich
and A. H. Ehrlich, Population, Resources, En-
vironment (Freeman, San Francisco, 1970),
chap. 7. More technical treatments of the
relationship between complexity and stability
may be found in R. H. MacArthur, Ecology
36, 533 (1955); D. R. Margalef, Gen. Syst. 3,
3671 (1958); E. G. Leigh, Jr., Proc. Nat.
Acad. Sci. U.S. 53, 777 (1965); and 0. T.
Loucks, "Evolution of diversity, efficiency,
and stability of a community," paper delivered
at AAAS meeting, Dallas, Texas, 30 Dec. 1968.
4. The figures used in this paragraph are all
based on data in Statistical Abstract of the
United States 1970 (U.S. Department of Com-
merce) (Government Printing Office, Washing-
ton, D.C., 1970).
5. A dramatic example of this effect is given
in R. Payne's analysis of the whale fisheries
[N.Y. Zool. Soc. Newsl. (Nov. 1968)]. The
graphs in Payne's paper are reproduced in
Ehrlich and Ehrlich (3).
6. If N is the number of people, then the numn-
ber of links is N(N-1)/2, and the number of
links per capita is (N-1)/2.
7. These figures and the others in this paragraph
are from Cleaning Our Environment: The
Chemical Basis for Action (American Chemi-
cal Society, Washington, D.C., 1969), pp.
95-162.
8. In his unpublished testimony before the
President's Commission on Population Growth
and the American Future (17 Nov. 1970),
Commoner acknowledged the operation of
diminishing returns, threshold effects, and so
on. Since such factors apparently do not ac-
count for all of the increase in per capita
impact on the environment in recent decades,
however, Commoner drew the unwarranted
conclusion that they are negligible.
9. R. Dumont and B. Rosier, The Hungry Fu-
ture (Praeger, New York, 1969), pp. 34-35.
10. L. Brown, Sci. Amer. 223, 160 (1970); P. R.
Ehrlich, War on Hunger 4, 1 (1970).
11. These figures are based on data from the
United Nations Statistical Yearbook 1969
(United Nations, New York, 1969), with
estimates added for the consumption by Main-
land China when none were included.
12. The notion that dispersed resources, because
they have not left the planet, are still avail-
able to us, and the hope that mineral supplies
can be extended indefinitely by the applica-
tion of vast amounts of energy to common
rock have been the subject of lively debate
elsewhere. See, for example, the articles by
P. Cloud, T. Lovering, A. Weinberg,
Texas Quart. 11, 103, 127, 90 (Summer 1968);
and Resources and Man (National Academy
of Sciences) (Freeman, San Francisco, 1969).
While the pessimists seem to have had the
better of this argument, the entire matter is
SCIENCE, VOL. 171
 academic in the context of the rate problem
we face in the next 30 years. Over that time
period, at least, cost, lead time, and logistics
will see to it that industrial economies and
dreams of development stand or fall with the
availability of high-grade resources.
13. B. Wattenberg, New Republic 162, 18 (4 Apr.
and 11 Apr. 1970).
14. These figures are from (11), from the FAO
Trade Yearbook, the FAO Production Year-
book (United Nations, New York, 1968), and
from G. Borgstrom, Too Many (Collier-
Macmillan, Toronto, Ont., 1969).
15. Man's Impact on the Global Environment,
Report of the Study of Critical Environmental
Problems (M.I.T. Press, Cambridge, Mass.,
1970).
16. A Model of Society, Progress Report of the
Environmental Systems Group (Univ. of
Signal Detectability
and Medical Decision-Making
Signal detectability studies help radiologists evaluate
equipment systems and performance of assistants.
Signal detection theory can be used
to investigate two problems of interest
to radiologists. First, the central con-
cern in the study of radiographic image
quality is to gain knowledge of the way
in which physical image quality affects
a diagnosis, not necessarily to design
high fidelity imaging systems (1). Sec-
ond, the increasing demand for diagnos-
tic radiology examinations has stimu-
lated studies to determine whether the
effectiveness and efficiency of radi-
ologists can be increased by the use of
trained technical assistants.
Detection theory is a basis for treat-
ing discrimination experiments in psy-
chophysics. In such experiments, one
attempts to learn something about a
sensory system by determining just how
small a change in some aspect of the
stimulus can be reliably detected. A
central feature of this analysis is the
distinction made between the criterion
that the observer uses to decide whether
The author is professor of radiology at the
University of Chicago, Chicago, Illinois 60637.
26 MARCH 1971
California Institute of Ecology, Davis, April
1969).
17. J. J. Spengler, in Population: The Vital
Revolution, R. Freedman, Ed. (Doubleday,
New York, 1964), p. 67.
18. R. Chrisman, Scanlan's 1, 46 (August 1970).
19. A more extensive discussion of this point is
given in an article by P. R. Ehrlich and A.
H. Ehrlich, in Global Ecology: Readings
Toward a Rational Strategy for Man, J. P.
Holdren and P. R. Ehrlich, Eds. (Harcourt,
Brace, Jovanovich, New York, in press).
20. J. L. Freedman, A. Levy, J. Price, R.
Welte, M. Katz, P. R. Ehrlich, in preparation.
21. J. Lederberg, Washington Post (15 Mar. and
22 Mar. 1970).
22. C. Smith, D. Simpson, E. Bowen, I. Zlotnik,
Lancet 1967-1, 1119, 1128 (1967).
23. Associated Press wire service, 2 Feb. 1970.
Lee B. Lusted
a signal is present and his sensory
capabilities as a signal detector. Re-
ceiver operating characteristic (ROC)
curves can be used to separate the sen-
sory and nonsensory variables. A large
body of literature is available on signal
detection theory in psychophysics (2)
and the use of ROC curves (3).
ROC Curve for Interpreting
Chest Roentgenograms
In 1946 a group of radiologists and
phthisiologists began an investigation to
evaluate the effectiveness of various
roentgenographic and photofluoro-
graphic techniques in detecting active
pulmonary tuberculosis. Yerushalmy,
who helped to initiate the study, has
recently reviewed the results and the
studies which followed (4). In the
course of the investigation it was dis-
covered that the variation in the in-
terpretations of chest roentgenograms
was of a disturbing magnitude: a phy-
sician would disagree with the diagnosis
24. P. R. Ehrlich and J. P. Holdren, BioScience
19, 1065 (1969).
25. See L. Brown [Seeds of Change (Praeger,
New York, 1970)] for a discussion of unem-
ployment problems exacerbated by the Green
Revolution.
26. G. Woodwell, Science 168, 429 (1970).
27. New York Times, 22 Oct. 1970, p. 18; News-
week 76, 50 (2 Nov. 1970).
28. S. S. Epstein, Environment 12, No. 7, 2
(Sept. 1970); New York Times service, 17
Nov. 1970.
29. G. S. Krantz, Amer. Sci. 58, 164 (Mar.-Apr.
1970).
30. R. A. Bryson and W. M. Wendland, in
Global Effects of Environmental Pollution,
S. F. Singer, Ed. (Springer-Verlag, New York,
1970).
31. J. Platt, Science 166, 1115 (1969).
of a colleague on an average of one
out of three times; on a second, in-
dependent reading of the same series
of chest films, a physician would dis-
agree with his own previous diagnosis
on an average of one out of five times.
The results of the intensive studies of
this phenomenon, which came to be
known as observer error, are shown in
an ROC graph in Fig. 1. The ROC
curve is plotted on normal-normal
coordinates (codex 41,453), according
to the detection theory convention of
false positive and true positive diagnoses
on the x- and y-axes, respectively. Two
parameters are abstracted from an
ROC curve: the slope, and the sensitiv-
ity index de', where de' is defined as
twice the normal deviates of the inter-
section of the ROC curve and the
negative diagonal. The slope is inter-
preted as the ratio of the standard
deviations of two distributions that,
hypothetically, underlie the detection
process. The measure d,' is normalized
by averaging the two variances of the
underlying data-generating distributions.
The more sensitively the observer per-
forms as a signal detector, the larger
the value of d'.
The ROC curve in Fig. 1 can ex-
plain the variation in roentgenogram
interpretation. Suppose that the six
points on the curve represent the diag-
noses of six different physicians who
have identical sensory capabilities for
detecting the signals (film densities) of
tuberculosis on the chest roentgeno-
gram, but they have different criteria
for what densities should actually be
called tuberculosis. One assumes that
they have the same sensory capabilities
because the index of detectability, de',
is the same for each physician.
The upper points on the curve repre-
sent individuals with more liberal
decision criteria, whereas the lower
1217